Keeping up with an industry as fast-moving as\u00a0AI<\/a>\u00a0is a tall order. So until an AI can do it for you, here\u2019s a handy roundup of recent stories in the world of machine learning, along with notable research and experiments we didn\u2019t cover on their own.<\/p>\n This week in AI, OpenAI signed up its first higher education customer: Arizona State University.<\/p>\n ASU will collaborate<\/a> with OpenAI to bring ChatGPT, OpenAI\u2019s AI-powered chatbot, to the university\u2019s researchers, staff and faculty \u2014 running an open challenge in February to invite faculty and staff to submit ideas for ways to use ChatGPT.<\/p>\n The OpenAI-ASU deal illustrates the shifting opinions around AI in education as the tech advances faster than curriculums can keep up. Last summer, schools and colleges\u00a0rushed<\/a>\u00a0to ban ChatGPT over plagiarism and misinformation fears. Since then, some have\u00a0reversed<\/a> their bans, while others have begun hosting workshops on GenAI tools and their potential for learning.<\/p>\n The debate over the role of GenAI in education isn\u2019t likely to be settled anytime soon. But \u2014 for what it\u2019s worth \u2014 I find myself increasingly in the camp of supporters.<\/p>\n Yes, GenAI is a poor summarizer<\/a>. It\u2019s biased and toxic<\/a>. It makes stuff up<\/a>. But it can also be used for good.<\/p>\n Consider how a tool like ChatGPT might help students struggling with a homework assignment. It could explain a math problem step-by-step or generate an essay outline. Or it could surface the answer to a question that\u2019d take far longer to Google.<\/p>\n Now, there\u2019s reasonable concerns over cheating \u2014 or at least what might be considered cheating within the confines of today\u2019s curriculums. I\u2019ve anecdotally heard of students, particularly students in college, using ChatGPT to write large chunks of papers and essay questions on take-home tests.<\/p>\n This isn\u2019t a new problem \u2014 paid essay-writing services have been around for ages. But ChatGPT dramatically lowers the barrier to entry, some educators argue.<\/p>\n There\u2019s evidence<\/a> to suggest that these fears are overblown. But setting that aside for a moment, I say we step back and consider what drives students to cheat in the first place. Students are often rewarded for grades, not effort or understanding. The incentive structure\u2019s warped. Is it any wonder, then, that kids view school assignments as boxes to check rather than opportunities to learn?<\/p>\n So let students have GenAI \u2014 and let educators pilot ways to leverage this new tech to reach students where they are. I don\u2019t have much hope for drastic education reform. But perhaps GenAI will serve as a launchpad for lesson plans that get kids excited about subjects they never would\u2019ve explored previously.<\/p>\n Here are some other AI stories of note from the past few days:<\/p>\n Microsoft\u2019s reading tutor:<\/a> <\/strong>Microsoft this week made Reading Coach, its AI tool that provides learners with personalized reading practice, available<\/a>\u00a0at no cost to anyone with a Microsoft account.<\/p>\n Algorithmic transparency in music:<\/strong><\/a> EU regulators are calling for laws to force greater algorithmic transparency from music streaming platforms. They also want to tackle AI-generated music \u2014 and deepfakes.<\/p>\n NASA\u2019s robots:<\/a><\/strong> NASA recently showed off a self-assembling robotic structure that, Devin writes, might just become a crucial part of moving off-planet.<\/p>\n Samsung Galaxy, now AI-powered:<\/a> <\/strong>At Samsung\u2019s Galaxy S24 launch event, the company pitched the various ways that AI could improve the smartphone experience, including through live translation for calls, suggested replies<\/a> and actions<\/a> and a new way to Google search using gestures<\/a>.<\/p>\n DeepMind\u2019s geometry solver:<\/strong><\/a> DeepMind, the Google AI R&D lab, this week unveiled AlphaGeometry, an AI system that the lab claims can solve as many geometry problems as the average International Mathematical Olympiad gold medalist.<\/p>\n OpenAI and crowdsourcing:<\/strong><\/a> In other OpenAI news, the startup is forming a new team, Collective Alignment, to implement ideas from the public about how to ensure its future AI models \u201calign to the values of humanity.\u201d At the same time, it\u2019s changing its policy<\/a> to allow military applications of its tech. (Talk about mixed messaging.)<\/p>\n A Pro plan for Copilot:<\/strong><\/a> Microsoft has launched a consumer-focused paid plan for Copilot, the umbrella brand for its portfolio of AI-powered, content-generating technologies, and loosened the eligibility requirements for enterprise-level Copilot offerings. It\u2019s also launched new features for free users, including a Copilot smartphone app.<\/p>\n Deceptive models: <\/strong><\/a>Most humans learn the skill of deceiving other humans. So can AI models learn the same? Yes, the answer seems \u2014 and terrifyingly, they\u2019re exceptionally good at it. according to a new study from AI startup Anthropic.<\/p>\n Tesla\u2019s staged robotics demo:<\/a> <\/strong>Elon Musk\u2019s Optimus humanoid robot from Tesla is doing more stuff \u2014 this time folding a t-shirt on a table in a development facility. But as it turns out, the robot\u2019s anything but autonomous at the present stage.<\/p>\n One of the things holding back broader applications of things like AI-powered satellite analysis is the necessity of training models to recognize what may be a fairly esoteric shape or concept. Identifying the outline of a building: easy. Identifying debris fields after flooding: not so easy! Swiss researchers at EPFL are hoping to make it easier to do this with a program they call METEOR.<\/a><\/p>\n Image Credits:<\/strong> EPFL<\/p>\n<\/div>\n \u201cThe problem in environmental science is that it\u2019s often impossible to obtain a big enough dataset to train AI programs for our research needs,\u201d said Marc Ru\u00dfwurm, one of the project\u2019s leaders. Their new structure for training allows a recognition algorithm to be trained for a new task with just four or five representative images. The results are comparable to models trained on far more data. Their plan is to graduate the system from lab to product with a UI for ordinary people (that is to say, non-AI-specialist researchers) to use it. You can read the paper they published here<\/a>.<\/p>\n Going the other direction \u2014 creating imagery \u2014 is a field of intense research, since doing it efficiently could reduce the computation load for generative AI platforms. The most common method is called diffusion, which gradually refines a pure noise source into a target image. Los Alamos National Lab has a new approach they call Blackout Diffusion<\/a>, which instead starts from a pure black image.<\/p>\n That removes the need for noise to begin with, but the real advance is in the framework taking place in \u201cdiscrete spaces\u201d rather than continuous, greatly reducing the computational load. They say it performs well, and at lower cost, but it\u2019s definitely far from wide release. I\u2019m not qualified to evaluate the effectiveness of this approach (the math is far beyond me) but national labs don\u2019t tend to hype up something like this without reason. I\u2019ll ask the researchers for more info.<\/p>\n AI models are sprouting up all over the natural sciences, where their ability to sift signal out of noise both produces new insights and saves money on grad student data entry hours.<\/p>\n Australia is applying<\/a> Pano AI\u2019s wildfire detection tech<\/a> to its \u201cGreen Triangle,\u201d a major forestry region. Love to see startups being put to use like this \u2014 not only could it help prevent fires, but it produces valuable data for forestry and natural resource authorities. Every minute counts with wildfires (or bushfires, as they call them down there), so early notifications could be the difference between tens and thousands of acres of damage.<\/p>\n Permafrost reduction as measured by the old model, left, and the new model, right.<\/p>\n<\/div>\n Los Alamos gets a second mention (I just realized as I go over my notes) since they\u2019re also working on a new AI model for estimating the decline of permafrost<\/a>. Existing models for this have a low resolution, predicting permafrost levels in chunks about 1\/3 of a square mile. That\u2019s certainly useful, but with more detail you get less misleading results for areas that might look like 100% permafrost at the larger scale but are clearly less than that when you look closer. As climate change progresses, these measurements need to be exact!<\/p>\n Biologists are finding interesting ways to test and use AI or AI-adjacent models in the many sub-fields of that domain. At a recent conference written up by my pals at GeekWire<\/a>, tools to track zebras, insects, even individual cells were being shown off in poster sessions.<\/p>\n And on the physics side and chemistry side, Argonne NL researchers are looking at how best to package hydrogen for use as fuel. Free hydrogen is notoriously difficult to contain and control, so binding it to a special helper molecule keeps it tame. The problem is hydrogen binds to pretty much everything, so there are billions and billions of possibilities for helper molecules. But sorting through huge sets of data is a machine learning specialty.<\/p>\n \u201c\u201cWe were looking for organic liquid molecules that hold on to hydrogen for a long time, but not so strongly that they could not be easily removed on demand,\u201d said the project\u2019s Hassan Harb. Their system sorted through 160 billion molecules<\/a>, and by using an AI screening method they were able to look through 3 million a second \u2014 so the whole final process took about half a day. (Of course, they were using quite a large supercomputer.) They identified 41 of the best candidates, which is a piddling number for the experimental crew to test in the lab. Hopefully they find something useful \u2014 I don\u2019t want to have to deal with hydrogen leaks in my next car.<\/p>\nMore machine learnings<\/h2>\n
![\"\"](\"https:\/\/techcrunch.com\/wp-content\/uploads\/2024\/01\/blackout-diffusion.jpg\")
<\/p>\n![\"\"](\"https:\/\/techcrunch.com\/wp-content\/uploads\/2024\/01\/permafrost-lanl.jpg\")
<\/p>\n